Truncation Strategies for Optimal Krylov Subspace Methods

Optimal Krylov subspace methods like GMRES and GCR have to compute an orthogonal basis for the entire Krylov subspace to compute the minimal residual approximation to the solution. Therefore, when the number of iterations becomes large, the amount of work and the storage requirements become excessive. In practice one has to limit the resources. The most obvious ways to do this are to restart GMRES after some number of iterations and to keep only some number of the most recent vectors in GCR. This may lead to very poor convergence and even stagnation. Therefore, we will describe a method that reveals which subspaces of the Krylov space were important for convergence thus far and exactly how important they are. This information is then used to select which subspace to keep for orthogonalizing future search directions. Numerical results indicate this to be a very effective strategy.

[1]  J. Meijerink,et al.  An iterative solution method for linear systems of which the coefficient matrix is a symmetric -matrix , 1977 .

[2]  J. Meijerink,et al.  Guidelines for the usage of incomplete decompositions in solving sets of linear equations as they occur in practical problems , 1981 .

[3]  S. Eisenstat,et al.  Variational Iterative Methods for Nonsymmetric Systems of Linear Equations , 1983 .

[4]  Gene H. Golub,et al.  Matrix computations , 1983 .

[5]  P. C. Robinson,et al.  A numerical study of various algorithms related to the preconditioned conjugate gradient method , 1985 .

[6]  Y. Saad,et al.  GMRES: a generalized minimal residual algorithm for solving nonsymmetric linear systems , 1986 .

[7]  Charles R. Johnson,et al.  Topics in Matrix Analysis , 1991 .

[8]  R. Freund,et al.  QMR: a quasi-minimal residual method for non-Hermitian linear systems , 1991 .

[9]  Henk A. van der Vorst,et al.  Bi-CGSTAB: A Fast and Smoothly Converging Variant of Bi-CG for the Solution of Nonsymmetric Linear Systems , 1992, SIAM J. Sci. Comput..

[10]  E. Desturler,et al.  Nested Krylov methods and preserving the orthogonality , 1993 .

[11]  Yousef Saad,et al.  A Flexible Inner-Outer Preconditioned GMRES Algorithm , 1993, SIAM J. Sci. Comput..

[12]  H. V. D. Vorst,et al.  The superlinear convergence behaviour of GMRES , 1993 .

[13]  Roland W. Freund,et al.  A Transpose-Free Quasi-Minimal Residual Algorithm for Non-Hermitian Linear Systems , 1993, SIAM J. Sci. Comput..

[14]  Cornelis Vuik,et al.  GMRESR: a family of nested GMRES methods , 1994, Numer. Linear Algebra Appl..

[15]  Joke Blom,et al.  VLUGR3: a vectorizable adaptive grid solver for PDEs in 3D, Part I: algorithmic aspects and applications , 1994 .

[16]  E. De Sturler.  Iterative methods on distributed memory computers , 1994 .

[17]  Ronald B. Morgan,et al.  A Restarted GMRES Method Augmented with Eigenvectors , 1995, SIAM J. Matrix Anal. Appl..

[18]  E. Sturler,et al.  Nested Krylov methods based on GCR , 1996 .

[19]  Anne Greenbaum,et al.  Any Nonincreasing Convergence Curve is Possible for GMRES , 1996, SIAM J. Matrix Anal. Appl..

[20]  Jan G. Verwer,et al.  Algorithm 758: VLUGR2: a vectorizable adaptive-grid solver for PDEs in 2D , 1996, TOMS.

[21]  Y. Saad,et al.  Deflated and Augmented Krylov Subspace Techniques , 1997 .

[22]  Gene H. Golub,et al.  Adaptively Preconditioned GMRES Algorithms , 1998, SIAM J. Sci. Comput..